unilm
unilm copied to clipboard
LayoutReader's very slow inference on CPU
Hello,
I am trying to use LayoutReader for post processing OCR results on a (decent) CPU device, but even after 10 minutes it fails to process even one instance! Is it expected or I am probably doing something wrong?
Thank you!
@siamakzd How many words are there in your OCR input?
Hi @wolfshow any update regarding inference speed? the model is quite slow even using a GPU .
Hi @wolfshow, is there anything new regarding the LayoutReaders speed? Still seems an unfixed bug.